A method for on-device learning of a machine learning network of an autonomous vehicle through multi-stage learning using an adaptive hyperparameter set, and an on-device learning device using the same

摘要

The present invention relates to a method for on-device learning of a machine learning network through multi-stage learning using adaptive hyperparameters. Classifying into n-th stage learning, generating first-stage learning data to n-th stage learning data, generating a first hyperparameter set candidate to hyperparameter set candidate based on default values of each of the hyperparameters, and the machine learning training a network, selecting the machine learning network with the highest performance, and generating a first adaptive hyperparameter set; (b) (k_1)th hyperparameter set candidates to (k_h)th hyperparameter set candidates are generated, trained using the kth stage training data, and the (k-1)th stage trained machine with the highest performance selecting a learning network and generating a k-th adaptive hyperparameter set; and (c) generating an n-th adaptive hyperparameter set and completing the current learning by performing n-th stage learning. A method and an apparatus using the same are disclosed.